WBCHSE Applied AI key terms

The field of building computer systems that perform tasks normally needing human intelligence, such as recognising images, understanding language, solving problems and making decisions from data.

An AI system designed for one specific task or a small set of tasks, such as translating text or recommending videos. It cannot transfer its skill to unrelated problems the way a person can.

A thought experiment proposed by Alan Turing in which a machine is judged intelligent if a human judge cannot reliably tell its text replies apart from a person's.

A practical rule of thumb that gives a good-enough answer quickly instead of a guaranteed best one. Search and planning programs use heuristics to cut down the options they examine.

Systematic unfairness in a model's outputs, usually learned from training data that under-represents or misrepresents some groups, or from choices made by the people who built the system.

The right and practice of controlling who collects personal data, what it is used for, how long it is kept and who it is shared with, including the safeguards that protect it.

The degree to which a person can understand why an AI system produced a particular output, which matters when decisions affect people's rights, safety, health or money.

The gap between people who have reliable access to devices, internet and digital skills and those who do not, which can widen inequality when services and jobs move online.

A named place in a program's memory that stores a value, such as a number or some text, and whose contents can be read or changed while the program runs.

A programming structure that repeats a block of instructions while a condition holds or for a fixed number of times, saving you from writing the same lines again.

An instruction such as if, elif or else that runs a block of code only when a test is true, letting a program choose between different paths.

A named, reusable block of code that takes inputs, does a defined job and may return a result. Functions keep programs shorter, easier to test and easier to read.

The process of finding, understanding and fixing errors in a program by reading messages, printing intermediate values, testing small parts and re-running the code.

An organised collection of related data, usually arranged in rows and columns or as files, used for analysis, visualisation or training a machine learning model.

Fixing or removing inaccurate, duplicate, incomplete or badly formatted records before analysis, so that results and models are not distorted by poor-quality input.

A value that lies far away from the rest of the data. It may be a genuine rare case or a recording mistake, so it should be checked before it is deleted.

Presenting data as charts, graphs or maps so that patterns, trends and comparisons become easy to see and explain, rather than being buried in tables of numbers.

Training a model on labelled examples, where each input is paired with the correct answer, so it can predict the label of new, unseen inputs.

Training a model on data without labels, looking for structure such as groups of similar records or fewer underlying factors that describe the data.

The labelled or unlabelled examples a model learns from. Its size, balance and accuracy strongly shape how well the finished model performs on new data.

When a model learns the training examples too closely, including their noise, and so performs well on them but poorly on new data it has not seen.

A measurable property of an example that a model uses as input, such as age, pixel brightness or word count. Good features make learning easier.

A portion of data held back from training and used only to measure how well a finished model performs on examples it has never seen.

The branch of AI that helps computers process human language, including tasks such as translating, summarising, answering questions and classifying the sentiment of text.

Splitting text into small units, such as words, sub-words or characters, that a language model can process and count. It is usually the first step in text analysis.

Using a model to decide whether a piece of text expresses a positive, negative or neutral attitude, often applied to reviews, survey answers and social media posts.

The branch of AI that lets computers interpret images and video, supporting tasks such as detecting objects, reading text, recognising faces and inspecting products on a line.

A system that suggests items such as products, films or courses by comparing a user's past behaviour and preferences with patterns found across many other users.

Using models to flag transactions or claims that look unusual compared with normal behaviour, so human reviewers can investigate them before money is lost.

Applying sensors, drones, satellite images and AI models to farm fields so that water, fertiliser and pesticide are used only where and when they are needed.

A model trained on very large amounts of text to predict and generate language. It can answer questions, draft text and translate, but may produce confident errors.

AI systems that create new content, such as text, images, audio or code, by learning patterns from existing examples rather than only labelling or predicting numbers.

A defined way for one program to request data or services from another, for example sending text to a translation model and receiving the translated result back.

Software whose source code is published, usually freely, so anyone can use, inspect and modify it. Libraries such as NumPy or scikit-learn save developers from writing everything themselves.

An interactive document that mixes runnable code, its output, formatted notes and charts in one place, widely used for exploring data and sharing analysis.

A plain-text file where each line is a record and values are separated by commas, commonly used to move tables of data between spreadsheets and programs.

A two-dimensional table structure in the pandas library with labelled rows and columns, offering quick ways to filter, group, merge and summarise large datasets in Python.

A short, specific description of the user, their need and the gap to be closed, which keeps a project focused and makes success measurable.

An early, simplified version of a product built to test an idea with real users and gather feedback before investing time and money in full development.

The smallest version of a product that still delivers real value to users, released early so the team can learn what works before building more features.